This repository contains minimal code for running the adaptive hybrid rational-LASG algorithm from the paper
P. Huwiler, D. Pradovera, and J. Schiffmann, Plug-and-play adaptive surrogate modeling of parametric nonlinear dynamical systems in frequency domain, International Journal for Numerical Methods in Engineering, 2023
Published version here. Preprint publicly available here.
The results in the above paper have been obtained by applying the code in this repository to a frequency-domain gas bearing model, which is not included in this repository. For more details on the bearing model, we refer to
E. Guenat and J. Schiffmann, Effects of humid air on aerodynamic journal bearings, Tribology International, 2018
Published version here. Preprint publicly available here.
- MATLAB® (coded and tested on version R2023a)
Sample scripts for surrogate training and testing are in tutorial_train_*.m and tutorial_test_*.m, respectively. The suffixes 1p1 and 1p2 refer to the number of parameters, 1 and 2 (in addition to frequency), respectively.
The target QoIs are defined in the function scripts getSampleImpedance_*.m (QoIs depending on both frequency and parameters) and getSampleForce_*.m (QoIs depending only on parameters). They can be customized to compute any desired QoI. The frequency and parameter ranges are defined near the beginning of tutorial_train_*.m, and can be easily adapted to new settings.